Unraveling the mysteries of the Green Fluorescent Protein

Vinod Subramaniam & T. M. Jovin

(Abteilung Molekulare Biologie)

The Green Fluorescent Protein (GFP) of the bioluminescent jellyfish Aequoria is revolutionizing molecular and cellular biology by virtue of its universal use as a noninvasive fluorescent marker. The ability to fuse GFP to proteins of interest and to express the constructs in vivo enables the visualization and tracking of proteins and complex intracellular processes by standard microscopy techniques. However, our understanding of the rich palette of physico-chemical properties of GFP and its mutants has not kept pace with the development of new mutants exhibiting different spectral properties. GFPs have been engineered to absorb light from 350-514 nm, and emit in the range of 400-530 nm. Photostability, fluorescence intensity, spectral peaks, pH sensitivity, emission quantum yields, extinction coefficients, have all been tuned by appropriate mutational strategies. The rational design and application of GFPs as indicators of structure and function requires a thorough understanding of their photophysical properties. These are very complex, including multiple lifetimes, blinking, switching, dark states, photobleaching, and triplet effects, which continue to confound researchers in the field. The full repertoire of biophysical techniques (steady-state and time-resolved fluorescence, NMR, ultrafast pump probe, single molecule spectroscopy, fluorescence correlation spectroscopy) is being used to solve the GFP puzzle, including the efforts of several groups in the Institute [AG Experimentelle Biophysik (Petra Schwille); Abt. Molekulare Biologie (Vinod Subramaniam, George Striker, and Thomas Jovin); AG Biomolekulare und chemische Dynamik (Peter Vöhringer); AG Einzelmolekulefluoreszenzspektroskopie (Claus Seidel)], while several other groups use GFP and its constructs for visualization of complex intracellular processes.

We have approached the problem of GFP photophysics in a few different ways, including steady-state fluorescence spectroscopy, high-resolution time-resolved fluorescence methods (1), two-photon excitation (2), fluorescence correlation spectro-scopy and high-resolution spectral hole-burning spectroscopy (3) at cryogenic temperatures. This last technique is a collaboration with the group of Prof. Silvia Völker at the Rijksuniversiteit Leiden, and the results of this work are the focus of this contribution.

Figure 1: Principle of hole-burning spectroscopy
A Schematic description of inhomogeneous broadening. The top panel contains depicts the absorption spectrum of a single molecule, with a characteristic (homogeneous) absorption linewidth G_hom. The next panel depicts the spectra of 10 molecules; each showing slightly different absorption line centers due to differences in their interactions with the host matrix. The next two panels depicts the case for 10³ and 10⁶ molecules respectively. In the latter case, the total absorption spectrum is a sum of the individual spectra for each molecule and is characteristically (inhomogeneously) broadened with linewidth -G_inhom. B Schematic description of an inhomogeneously broadened absorption line (center wavelength l₀) and the hole burned at the wavelength of laser excitation (l₁). The photoproduct (absorbing at a different wavelength) is also depicted. C (Left Panel): Excitation spectrum of wild-type GFP at 1.6 K before (æ) and after (---) burning a hole into A at 434±1 nm. The insets in the top left and bottom right of the panel are typical holes burned into the A and I states respectively. The hole widths ~ 2 cm ^-1 are limited by the laser bandwidth. The fundamental absorption transitions are determined to be A_0-0 = 434±1 nm; I_0-0 = 495±1 nm. (Right Panel): Corresponding emission spectrum of I upon direct excitation of I at 493 nm. The I state is produced upon burning into A, clearly demonstrating the photoconversion A-> I. The GFP was dissolved in a buffer/glycerol solution that transforms into a glass at cryogenic temperatures. Data taken from (3).

The Technique: Hole Burning Spectroscopy
Spectral holeburning (HB) is a high resolution technique for determining the homogeneous linewidth (G_hom) of a transition hidden underneath the intrinsically inhomogeneously broadened absorption band (width G_inh) of a chromophore embedded in a host matrix (Figure 1A and B). The homogeneous linewidth yields information on population decay and energy-transfer processes as well as optical dephasing and spectral diffusion. In judicious combination with variation of physical parameters, holeburning can yield information about vibrational frequencies and relaxation times, local microscopic environments, and selective photochemistry. In practice, the absorption band of the chromophore is irradiated with a narrow-band cw laser. Molecules resonant with the laser line undergo a photochemical or photophysical transition leading to photoproducts absorbing at different wavelengths; the result is the creation of a hole at the laser wavelength in the absorption band. The burnt hole is subsequently probed by ei-ther transmission or fluorescence excitation spectroscopy by scanning the laser frequency over the chromophore absorption band (Figure 1C). Reviews of the holeburning technique and its applications to proteins can be found in refs (4, 5).

Holeburning and GFP
The GFP chromophore results from an autocatalytic cyclization and subsequent oxidation of the three residues Ser 65, Tyr 66 and Gly 67 (6-8). The chromophore is nestled within a cylindrical "b-can" structure and is very effectively shielded from the solvent (Figure 2A). The room-temperature absorption spectrum of wild-type GFP (wt-GFP) is characterized by two maxima attributed to different protonation states of the chromophore: the maximum at ~398 nm corresponding to a neutral A-form and that at ~478nm to an anionic or deprotonated B-form (9-13) (Figure 2B). The wild type GFP molecule has long been known to exhibit a photochromicity between the A and B forms (1, 7, 14), as well as a photoconversion postulated to involve excited-state proton transfer from the protonated A- to the deprotonated B-form, passing through an excited intermediate I-form, which has been considered to be responsible for the green fluorescence. This intermediate form had not been spectrally identified and the details of the photoconversion remained unclear until the holeburning studies identified the I-form of wild-type GFP in absorption (3). The holeburning experiments, in combination with absorption, excitation and emission spectroscopy have unravelled the photo-interconversion pathways.

To solve the pathways of photoconversion, it is necessary to

Figure 2: GFP Structure and Photophysics
A Ribbon diagram of the WT-GFP structure based on X-ray crystallography data of Yang et al. (17). The GFP exhibits a unique cage-like structure composed of a 'b-can' encompassing the chromophore (in green), and shielding it from the solvent. B Normalized fluorescence excitation and emission spectra (dashed red and solid blue lines respectively) for the wild-type GFP. The insets depict the protonated and deprotonated states of the chromophore (formed by cyclization and oxidation of Ser 65, Tyr 66 and Gly 67). The deprotonated chromophore is thought to be responsible for the secondary excitation peak at ~ 477 nm and also for the green emission. C Proposed energy level scheme for WT-GFP derived from holeburning, emission- and excitation-spectroscopy. The 0-0 transitions for the photointerconvertible A, I, and B states, as well as vibronic transitions are indicated. Data taken from (3). D Modified energy level scheme based on ESPT model (11,12) derived from time-resolved fluorescence data (2). The dashed lines between B* and I* represent the transition for the RSGFP mutant. This transition is not present in WT-GFP or S65T (see barrier in Figure 2C). High precision timeresolved data are able to resolve the I*-> I and B*-> B lifetimes as ~3.3 ns and ~2.8 ns respectively for WT-GFP (1).

Establishing the transition energies:

We found it impossible to burn narrow spectral holes at arbitrary positions in the absorption bands; effects were obtained in the vicinity of the expected 0-0 transitions (i.e., the transition from the lowest vibronic level of the ground state to the lowest vibronic level of the excited state). For example, by selective excitation into the A-form within the range 434 ± 1 nm, narrow holes could be burnt, from which we conclude that A_0-0 lies at 434 ± 1 nm. Similarly, the I_0-0transition and the B_0-0 transition were established at 495 ± 1 nm and at 477 ± 1 nm. Note that the observed hole widths (Figure 1C) were limited by twice the laser bandwidth of ~1 cm^-1 and thus are not representative of the homogeneous linewidth G^hom ; that is, they do not yield any information about dynamical processes.

Determining the relative positions of the states:

Having established the 0-0 transitions of the three states, close inspection of the absorption, excitation and emission spectra upon cooling to 1.6K and comparison with the room temperature spectra yielded further insights into the energy level scheme. Inversion of the ratio of absorbances of the A- and B-forms at 1.6 K with respect to that at 295 K indicated that B has a slightly lower ground-state energy than A. Further, the observation that the I-form is populated at 295 K but not at 1.6 K suggested that the ground state I energy must be somewhat higher than that of A and B, by a few hundred cm^-1. I, when populated by burning into either A or B at 1.6 K, remains populated upon warming to 150 K, but tends to the low thermal equilibrium value at 295 K, suggesting that the height of the barriers in the ground state separating A and B from I must also be of the order of hundreds of wavenumbers. This establishes the relative positions of the ground states. For details of the experiment, see ref (3).

Establishing barriers in the excitedstate, and interconversion mechanisms:

The low temperature emission and excitation spectra exhibit a strong variation with excitation and detection wavelengths, and point to the role played by the I-form. For example, for excitation below 435 nm both A and B forms are excited and the A*-> I* process occurs. For excitation above 435 nm, only the B-form is excited. A stable I form is produced by burning into A, as well as by burning into B. In the former case, the A*-> I*-> I process is most likely responsible. Burning in B, however, leads to no emission from I*, suggesting that production of I in the ground state occurs through the radiationless conversion B*-> I (see dashed arrows in Fig. 2C). In the excited state, the barrier for the process A*-> I* appears to be low, while in contrast, the barrier for B*-> I* should be at least 2000 cm^-1 since excitation into B*, even in its vibronic bands, does not induce any emission from I*, but only produces I in the ground state.

Implications for Applications:

These experiments demonstrate that reversible optical switching is not only revealed in single-molecule experiments (15) but can also be observed in an ensemble of GFP-molecules at low temperature. We have found similar results for red-shifted mutants of GFP (Creemers, et al., manuscript submitted). Although thermal equilibrium at room temperature strongly favours one of the three forms A, I or B in these mutants, this form can still be photo-interconverted into the others, and may have important consequences when using GFP-mutants as fusion tags or markers in cells and organisms. For example, an observed change in the emission spectrum arising from an intra-molecular photo-interconversion may be erroneously attributed to dynamic intermolecular processes occurring within the cell. In the study of protein-protein interactions by FRET, a change in the colour of the fluorescence is commonly interpreted as evidence for energy transfer and, therefore, for an interaction between proteins. However, such a colour change may also arise from a photoinduced conversion between conformers within a specific GFP-mutant, thus making it essential that one distinguishes between phenomena originating from intra- and intermolecular interactions.

Time-resolved Fluorescence of GFPs:

In collaboration with George Striker (Abt. 060) and Andreas Volkmer (formerly Abt. 010) we have also applied time-resolved fluorescence techniques with one- and two-photon excitation to further explore the spectroscopic characteristics of GFPs (1, 2). This work was performed in the Femtosecond Research Center at the University of Strathclyde, Glasgow, and in the laboratory of Claus Seidel.

Although the literature has reported predominantly monoexponential lifetimes (~3ns) for the wild-type and some mutant GFPs, using high-precision data we have established that the wild-type and two mutants (EYFP and RSGFP) clearly exhibit multiexponential decays, and distinguished the fluorescence lifetimes of the I and the B species of wild-type GFP (~3.3 and ~2.8ns respectively) (1). Wild-type GFP also exhibits a strong photochromicity between the A and B states upon 400 nm excitation, which does not alter the fluorescence lifetimes, but does modulate the amplitudes. This phenomenon was exploited to assign lifetimes to the various species.

We have also compared one- (OPE) and two-photon excitation (TPE) of the wild-type GFP and the S65T and RSGFP mutants (2). Both OPE and TPE yielded similar fluoresence decay times, rotational correlation times, and steady state emission spectra, suggesting that the fluorescence originated from the same excited singlet states. The RSGFP mutant exhibits complex decay kinetics with a dominant component of 1.1 ns lifetime. The fluorescence lifetimes combined with relative quantum efficiency calculations yields estimations of the radiative and non-radiative rate constants for S65T-GFP and RSGFP. We postulate that the excited electronic intermediate state I* may be distinguished by an excitation-history dependent environment associated with the chromophore, that is, upon whether I* is accessed via the A* or B* manifolds (Fig.2D). The terminology IA* and IB* denotes these substates and reflects this excitation history; additionally, each of these substates can have a (different) characteristic non-radiative rate, reflecting different internal environments of the excited state chromophore. This modified kinetic scheme is the simplest model consistent with our time-resolved fluorescence data for GFPs and with the high resolution time-resolved (1) and hole-burning (3) spectroscopy data upon OPE. We infer that the two longer lifetimes of RSGFP (1.1 ns and 3.4 ns) represent the deactivation of the two substates I_B* and I_A* respectively.

We note that some of the time-resolved data do not directly confirm the holeburning results; this may have to do with the gross differences in solvent composition and temperatures used in the holeburning and time-resolved studies.

Future Directions:

In the holeburning experiments, the linewidths of the burnt holes were limited by the laser bandwidth, and thus do not yield information about dynamic processes. A natural extension of this work is to explore the dynamics of the photointerconversion process by extracting the homogeneous linewidths of the molecules. Investigations of other GFP mutants will yield further insights into the role played by key residues in the protein.

So far mutations in GFP have yielded proteins with fluorescence excitation and emission spectra to the yellow-green part of the spectrum. Extension into the red part of the spectrum has proved elusive, and would have significant implications for imaging without the effect of autofluorescence as well as for additional energy transfer applications. A recent breakthrough has yielded fluorescent proteins from unrelated organisms, nonbioluminescent fluorescent corals of the genus Anthozoa, emitting in the yellow and red-orange regions of the spectrum (16). Although these proteins have very little sequence identity with GFPs, they preserve many of the structural features including the "b-can" fold, and some key residues in the chromophore and its surroundings. These proteins expand the palette of fluorescent proteins for multiwavelength imaging, and promise to be good acceptors for FRET from 'green' GFPs.

Acknowledgment:

This work is the result of a long-term collaboration with the group of Prof. Silvia Völker at the Rijksuniversiteit Leiden in the Netherlands. The spectroscopy experiments were carried out in Leiden by Tijsbert Creemers and Arjan Lock. We thank George Striker (Abt. 060) and Andreas Volkmer (formerly Abt. 010) for close collaborations and stimulating discussions.

References:
1. Striker, G., Subramaniam, V., Seidel, C. A. M., and Volkmer, A. (1999) J. Phys. Chem. B 103, 8612-8617.
2. Volkmer, A., Subramaniam, V., Birch, D. J. S., and Jovin, T. M. (1999) Biophys. J. in press.
3. Creemers, T., Lock, A., Subramaniam, V., Jovin, T., and Völker, S. (1999) Nature Struct. Biol. 6, 557-560.
4. Völker, S. (1989) Annu. Rev. Phys. Chem. 40, 499-530.
5. Friedrich, J. (1995) Meth. Enzym. 246, 226-259.
6. Heim, R., and Tsien, R. Y. (1996) Current Biology 6, 178-82.
7. Cubitt, A. B., Heim, R., Adams, S. R., Boyd, A. E., Gross, L. A., and Tsien, R. Y. (1995) Trends Biochem. Sci. 20, 448-455.
8. Niwa, H., Inouye, S., Hirano, T., Matsuno, T., Kojima, S., Kubota, M., Ohashi, M., and Tsuji, F. I. (1996) Proc. Natl. Acad. Sci. USA 93, 13617-22.
9. Ward, W. W., Cody, C. W., Hart, R. C., and Cormier, M. J. (1980) Photochem. Photobiol. 31, 611-615.
10. Palm, G. J., Zdanov, A., Gaitanaris, G. A., Stauber, R., Pavlakis, G. N., and Wlodawer, A. (1997) Nature Struct. Biol. 4, 361-365.
11. Chattoraj, M., King, B. A., Bublitz, G. U., and Boxer, S. G. (1996) Proc. Natl. Acad. Sci. USA 93, 8362-7.
12.Lossau, H., Kummer, A., Heinecke, R., Pöllinger-Dammer, F., Kompa, C., Bieser, G., T., J., Silva, C. M., Yang, M. M., Youvan, D. C., and Michel-Beyerle, M. E. (1996) Chem. Phys. 213, 1-16.
13. Brejc, K., Sixma, T. K., Kitts, P. A., Kain, S. R., Tsien, R. Y., Ormo, M., and Remington, S. J. (1997) Proc. Natl. Acad. Sci. USA 94, 2306-11.
14. Patterson, G. H., Knobel, S. M., Sharif, W. D., Kain, S. R., and Piston, D. W. (1997) Biophys. J. 73, 2782-90.
15. Dickson, R. M., Cubitt, A. B., Tsien, R. Y., and Moerner, W. E. (1997) Nature 388, 355-8.
16. Matz, M. V., Fradkov, A. F., Labas, Y. A., Savitsky, A. P., Zaraisky, A. G., Markelov, M. L., and Lukyanov, S. A. (1999) Nature Biotech. 17, 969-973.
17. Yang, F., Moss, L. G., and Phillips, G. N. (1996) Nature Biotech. 14, 1246-1251.